Mint/(1984) Vol.

XCIII, 161-176

Seeing the Present

J. BUTTERFIELD

1. Introduction
Philosophers of time fall into two broad camps. For detensers, the
present is an epistemological/subjective notion reflecting our

Downloaded from http://mind.oxfordjournals.org/ by guest on May 14, 2012

limited knowledge at any time of a temporally extended reality, just
as where we are limits our knowledge of distant objects (Taylor
'955'. Quine i960; Section 36; Mellor 1981). Tensers distrust this
analogy between the present and the local; for them the present but
not the local is an ontological/objective notion (Prior 1959, 1970;
Dummett i960; Gale 1968, p. 213). This crude contrast is made
more precise in various ways. For example, detensers accept the
idea of an actual past and future—what in fact has happened and
will happen; tensers usually reject the idea of an actual future, and
sometimes (Dummett 1968, Prior 1970) of an actual past.
I am a detenser. But there is no doubt that strong intuitions seem
to support the tenser's view that the present is objective. Here are
three. We more readily take as real the presently-existing objects,
wherever they are, than the objects that are at some time located
here (Putnam 1967; Lloyd 1978, p. 217). We are more apt to give
sentences time-variable truth-values than space-variable ones
(Dummett 1973, pp. 386, 390). And we think of ourselves as sharing
a common, albeit ever-changing, now, while we each have a
different here (Gale 1964, p. 105). If all and only presently-existing
objects (but no objects that are sometime here) are in some sense
real, the present is surely objective in a way the local is not.
Similarly the second and third intuitions suggest that the present
and its 'movement' are in some way objective.
Detensers must explain away such intuitions, i.e. explain why we
have them without endorsing them, or without conceding that they
support the tenser. This paper attempts to do that for just these
three intuitions. I restrict myself to these three not because they
exhaust the tenser's possibilities, or because other intuitions cannot
be disposed of;1 but because these three have a unified explanation
1 For some examples, cf. Taylor (1955), Garson (1971), Mellor (1981, Chaps. 3
and 5) and Butterfield (1984).
161
162 J. BUTTERFIELD:

which detensers have so far missed. The basis of the explanation is

that both in observation and communication time-lags are usually
short enough to be ignored. We can usually take observation to
inform us of the present state of distant objects, and communication
to inform us of the speaker's present belief. I argue this in Sections 2
and 3; I then explain the three intuitions in a way which does not
make the present objective.

2. Observing the Present

Observation takes time: the time-lag is due both to neural proces-

Downloaded from http://mind.oxfordjournals.org/ by guest on May 14, 2012

sing by the observer and, for sight, hearing and smell, to the
transmission of signals to our sense organs. We can however almost
always ignore this time-lag. That is, if the process of observation is
reliable, we can take ourselves to learn the observed object's state at
the time we make our observational judgment, and nor merely at
some previous time. We can judge 'a is now G' or 'a is G at t' where t
is the time of judgment; and not merely 'a was G' or 'a was G at t",
where t' is some previous time at which, we believe, signals left the
object. Similarly for ascriptions of relations: we can judge 'a is now
H to b' or 'a is H to b at t', though the time-lags involved in
observing a and b may be rather different, say because a is further
away than b.
That we can usually observe objects' present states has been
remarked before (Dummett 1973, p. 388; Mellor 1981, p. 157). But
without some estimates of the time-lags involved in observation the
claim may be dismissed as mere conjecture (Zemach 1972, p. 253).
(And I must admit that smell is an exception: we can smell burnt
toast long after the toast has stopped burning. I only make the claim
for the other senses: touch, sight and hearing.) My estimates will
concentrate on sight and on the ascription of properties rather than
relations; the other cases can be dealt with similarly.
We must of course estimate the distances of observed objects, the
transmission speeds of light signals, and the time we take to process
these signals into observational judgments. Most objects we
observe are less than 1000 metres away, and light travels this sort of
distance almost instantaneously. Since we process light entering
our eyes in about half a second, that is the typical time-lag in
observation. 1
1 Processing takes about half a second for many kinds of observation; cf. Posner
and Mitchell (1967), Potter and Faulconer (1975), Flowers (1975) and Shaffer
and LaBerge(ig79).
SEEING THE PRESENT 163

But we must also estimate how rapidly objects change their

observable properties. If these properties changed fast enough, an
object's closeness and the high speed and rapid processing of signals
from it would not justify our judging it to have an observed property
at the time of our judgment. In other words, we must estimate not
only the time-lag, but also a limit the time-lag must not exceed if we
are to be entitled to ignore it.
Estimating this limit needs some care. Call the shortest length of
time for which a can be G a's time-scale for G. Suppose for example
it is five seconds. Primafacie then, a judgment 'a is now G' (or 'a is G
at t', t the time of judgment) would need to be made within five

Downloaded from http://mind.oxfordjournals.org/ by guest on May 14, 2012

seconds of the light leaving a, if it is to be correct.
But this is not quite right. We must allow a's changing with
respect to G within the time-lag of observation. Suppose a does in
fact change with respect to G every five seconds, and .that photons
leading to an observation are equally likely to leave a at any time.
Then if a judgment 'a is now G' is made half a second after the
photons which cause it leave a, its probability of being true will be
9/10. For photons leaving a in the first 9/10 of a five-second period
during which a is G (i.e. the first \\ seconds) will lead to true
judgments; photons leaving in the last tenth (i.e. the last half-
second) will lead to false judgments. (This of course assumes that
our perception is reliable in other respects, i.e. that it will make us
judge a to be G if it was G when the light left it.) Similarly for other
time-scales: if a judgment 'a is now G' is to have a 90 per cent
probability of being true, the time-lag must be less than a tenth of
the time-scale.
But how then can we claim to observe the distant present? We
certainly think that fewer than 10 per cent of our observational
judgments are falsified by objects changing during the time-
lag in observation; yet given a time-lag of half a second, that means
a time-scale of at least five seconds. And surely most objects
can have most of their observable properties for less than five
seconds.
Indeed they can: but they don't. In the above calculation I
assumed that a changes with respect to G as often as it could change.
That frenetic assumption is of course generally false, and its falsity
allows us a longer time-lag for a given probability of truth. What we
need is the average time for which a does not in fact change with
respect to G, not the shortest time. Suppose for example that on
average a changes with respect to G every ten seconds. Then with a
164 J. BUTTERFIELD:

time-lag of half a second, judgments 'a is now G' get a probability of

95 per cent; and the less frequent the changes, the greater the
probability will be.
This makes all the difference. Most of the observable objects
around us are solids—'moderate-sized specimens of dry goods' as
Austin called them (1962, p. 8); and these actually change their
observable properties very infrequently. So time-lags very rarely
falsify our judgments about their observable properties at the time
we make the observation; falsity in our judgment almost always
arises for other reasons than a change in the object during the time-
lag.

Downloaded from http://mind.oxfordjournals.org/ by guest on May 14, 2012

Given that we can usually ignore observational time-lags, why
can we ? Why is the time-lag about half a second; and why do most
observable properties change rarely enough for that time-lag to be
ignored ? I think we should take the second question first: once we
have explained why observable properties change as rarely as they
do, we can expect there to be a functional (ultimately, evolutionary)
explanation of why the time-lag is about half a second. In outline,
the explanation would be that, provided perception is otherwise
reliable, such a time-lag allows us confidently to make present-
tensed, not just past-tensed, observational judgments. And that is
useful since our actions are more likely to succeed if guided by true
beliefs about the present, and not merely the past, state of the world;
our actions may still fail, but at least we avoid failure due to changes
occurring since the start of the process of perception.
Why then do most solid objects change their observable pro-
perties so rarely? Surely this is because of their resilience to the
frequent internal and external disturbances of various kinds that
bring about change, and the in frequency of major disturbances they
would not be resilient to. I consider in turn mechanical, electro-
magnetic, chemical and thermal disturbances. Most observable
objects are too massive and too strong to be changed by common
sorts of mechanical disturbance such as air pressure, and low-speed
impacts with other objects. They are nearly electrically neutral, and
so not seriously subject to electromagnetic forces. They are resilient
to most of the chemical reactions that could befall them, being
chemically stable so that reactions will not originate internally, and
massive enough to be little affected by most common chemical
interactions with the environment. Finally, they are not often liable
to be changed by internal thermal agitation, nor by external
heat.
SEEING THE PRESENT 165

3. Communicating with the Present

Communication takes time: in speech the time-lag is due to
neural processing by speaker and hearer, the time taken to speak,
and the time speech takes to reach the hearer. Similarly, of course,
for written communication. For speech we can almost always ignore
this time-lag: someone hearing the speaker assert that p can take
himself to have learnt the speaker's present belief that/) (supposing
the speaker to be sincere), not just a past one. That is, he can take the
speaker to believe that p at the time of reception of the message and
not just at the time of utterance,1 (p need not of course be a present

Downloaded from http://mind.oxfordjournals.org/ by guest on May 14, 2012

tense proposition; it can have any tense or be tenseless). As for
assertions, so for requests; the hearer can take himself to learn what
the speaker desires at the time of reception, not just what he desired
at the time of utterance. And similarly for other speech-acts.
These claims are justified as in Section 2: the time-lags in
communication are generally less than some allowable maximum.
In observation the maximum is set by (i) some prescribed prob-
ability of correctly judging an object's state at the time of judgment
and (ii) the frequency with which objects change their observable
properties. Here it is set by (i) a prescribed probability of the hearer
correctly judging what the speaker believes, desires etc. at the time
of reception, and (ii) the frequency with which speakers change the
beliefs, desires etc. that they communicate.
The time-lags in oral communication are generally longer than
those in observation: usually several seconds—and mostly due to
the time it takes to speak the utterance. Nevertheless they can
usually be ignored. The reason, as in the case of observation, is that
(ii) is so low that the time-lags stay below the maximum even if we
set (i) high. This is clear from our experience. We habitually say
what we believe or desire, confident that we will still believe or
desire it when the hearer understands our utterance. There are
exceptions (racing commentaries for instance!); but only in the
much slower process of written communication is it at all common
for us to change our minds before the message is received.
That time-lags in communication .can usually be ignored greatly
enhances an individual's ability to observe the distant present.

Allowing for insincerity: the hearer can take the speaker to intend him to think
that the speaker believes that p at the time of reception. I will ignore insincerity
throughout; the amendments needed to allow for it are obvious. (Cf. how I
ignored the unreliability of observation in Section 2.)
166 J. BUTTERFIELD:

When we hear an observational report 'a is now G' or 'a is G at «' (* is

time of utterance) we can usually take the report to claim that a is G
at the time of reception as well as at the time of utterance. And since
other observers are usually sincere in their observational reports,
and their perceptions as reliable as our own, we can usually believe
that a is G at the time of reception. The point is obviously important
when we have to coordinate with one another at a distance, e.g.
getting a crane hook into the right position to attach the load, or
moving a sofa upstairs.
Given that we can ignore time-lags in communication, one can
still ask, as in Section 2, why this is so. Why is the time-lag typically

Downloaded from http://mind.oxfordjournals.org/ by guest on May 14, 2012

several seconds? And why do the beliefs etc. that we communicate
change rarely enough for these time-lags to be ignored? My
argument does not need answers to these questions, but I should
emphasise that unlike the case of observation, we cannot expect to
explainers' why the beliefs etc. we communicate change rarely, and
then give a functional (ultimately, evolutionary) explanation of why
the time-lag is typically several seconds. A time-lag short enough to
be ignored would indeed be useful just as in Section 2, since our
actions are more likely to succeed if guided by true beliefs about the
speaker's present, and not merely past, beliefs. But we cannot
expect to treat the two questions separately. For it seems likely that
the time-lag in communication influences which beliefs etc. we
communicate. At least, such an influence, by for example limi-
tations of the vocal tract, seems more likely than the parallel in
observation: viz. that the observational time-lag influences what
counts as an observational property.

4. Restricting Existence to the Present

We are far more inclined to say that all and only present objects exist
than that all and only objects that are at some time here exist. At
first, it seems clear why there is this asymmetry. Both tensers and
detensers admit that 'All and only present objects exist' is ambigu-
ous; since objects are created and destroyed, we must distinguish
two senses of existence. There is the narrow sense, present
existence, in which Reagan exists and Caesar does not; and a wider
sense, timeless existence, in which all past present and future
objects exist.1 Obviously, 'All and only present objects exist' is true

1 If talk about alternative possible futures or pasts requires objects that never in
fact exist, there will be a still wider sense including such possibilia. Also, if one
SEEING THE PRESENT 167

or false, according as 'exist' means present or timeless existence.

But in any case 'exist' never means 'are at some time here'; hence the
asymmetry.
Detensers cannot however just stop there. Why does 'exist' never
mean 'are at some time here', when it sometimes means 'are
somewhere now' ? It is surely no accident that the narrow sense of
'exists' is present existence and not its spatial analogue; and the
reason for this asymmetry, whatever it is, may well undermine the
detenser's analogy between the present and the local. Moreover, it
is often suggested that present existence sould be analysed by the

Downloaded from http://mind.oxfordjournals.org/ by guest on May 14, 2012

existential quantifier; this means associating with each time a
domain of quantification containing the objects that then presently
exist, and using temporal operators to pass us from one time to
another as in tense logic. But no one has made the analogous
suggestion: that location here at some time should be analysed by
the quantifier, with spatial operators used to pass us from one place
to another. Why not?
Sections 2 and 3 above give the detenser answers to both these
questions. He can explain why the narrow sense of 'exists' is present
existence, but not being at some time here; and why present
existence (but not being at some time here) can be analysed by the
quantifier, without making the present objective. I consider these in
turn.
That 'exists' means present existence follows from its use as an
observational predicate which is ascribed to whatever can be
observed, and which can apply to an object at one time and not at
another. The first feature serves to contrast with 'is red' etc.; the
only properties 'exists' will demand of its instances will be
properties about location in space and time. The second feature
serves simply to set aside tenseless existence, i.e. being at some time
located somewhere. Since we can predicate 'exists' of distant
objects, it does not mean 'is now (or was or is at some time) located
here'. And since by Section 2 we can usually take observed objects
not to be destroyed during the observational time-lag, 'exists' does
not mean 'was located somewhere' or 'was located somewhere in my
field of observation'. Rather it means 'is now located somewhere' or
'is now located somewhere in my field of observation'. This last

supposes there is no actual future one will reject timeless existence, though one
will accept a wider sense including possibilia. But allowing for these compli-
cations will not affect the argument.
168 J. BUTTERFIELD:

possibility is ruled out by the fact that someone who is told that a
exists will not refuse to assert that a exists merely because he himself
cannot observe a.1
Sections 2 and 3 also explain why the quantifier can analyse
present existence, but not location here at some time. We have only
to assume that ceteris paribus we should analyse observational
sentences (i.e. sentences reporting observational judgments) by
simple rather than complicated formulae.
By Section 2, observation usually informs us of objects' prop-
erties and relations at one time (the time of judgment) though

Downloaded from http://mind.oxfordjournals.org/ by guest on May 14, 2012

they are not then at one place. Thus observational sentences
generally have a temporally local but spatially dispersed subject-
matter. The time can be specified by an adverbial expression
governing the sentence; while spatial information will be naturally
specified by adjectival expressions qualifying terms for the objects.
So temporal operators give observational sentences simpler analy-
ses than spatial operators do.
Furthermore, if the quantifier analyses present existence, present-
tensed observation sentences can be analysed by formulae without
temporal operators: 'a is now G' and 'Something is now H to b'
will be analysed by the formulae 'G(a)' and \3x)H(x,b)'. But if
the quantifier analyses location here at some time, i.e. we associate
with each place a domain containing the objects that are ever there,
such sentences will generally have to be analysed by formulae with
spatial operators. For only with operators can we construct
formulae mentioning objects in different domains. A sentence like
'Someone is taller than b' will thus get a complicated analysis, just
as in tense logic 'Someone (now dead) was taller than b' gets a
complicated analysis. Only if we wanted to give simple analyses to
sentences with a spatially local and temporally dispersed subject-
matter, such as 'It often rains here', would we use the quantifier to
analyse location here at some time. But we surely want to do no such
thing: such states of affairs as its often raining here seem to us
intrinsically complex—because they do not reveal themselves in
single observations. 2

1 This explanation carries over to other observational predicates. For example,

we can similarly explain why 'is red' means 'is now red and is located
somewhere'. J admit that some further anti-verificationist premiss is needed to
prevent e.g. taking 'exists' ('is red') to mean 'is now (red and) located
somewhere in someone's field of observation'. For a recent discussion, cf.
Williams (1981).
2 This argument against spatial operators does not require them to exclude
SEEING THE PRESENT 169

These considerations are strengthened by Section 3. For the

rapidity of oral communication gives us pragmatic reasons for using
observational sentences with a present-tense like 'a is now
G', rather than those with a date like 'a is G at t' or with another
tense like 'a was G'. If communication took long enough for a to
have a good chance of changing with respect to G, or if we
could not predict the time-lag, we might save our hearer having
to guess or work out the time for which we are asserting a to
be G by using a date: 'a is G at t.' This is of course what we
do by dating letters which take a long and unpredictable time
to arrive. If the time-lag were long but predictable, we might use

Downloaded from http://mind.oxfordjournals.org/ by guest on May 14, 2012

an indexical appropriate to the hearer's point of view: if we knew
communication would take five minutes, we might say 'Five
minutes ago, a was G.' 1 But as it is, oral communication is mostly
fast enough for us to mean to be heard to say 'a is now G', and to
be believed.
So much by way of explaining our tendency to prefer a concept of
present to one of spatially local existence, and to analyse it by the
existential quantifier. I must emphasise that in arguing against a
quantifier analysis of location here at some time, and against spatial
operators, I am not arguing for the quantifier analysis of present
existence, nor for temporal operators—both commonly advocated
by tensers.
First, we can give present-tensed observational sentences simple
analyses without using the quantifier to analyse present existence, or
using temporal operators. Wilson (1955), for instance, suggests
placing all past, present and future objects in a single domain,
together with times which can be referred to by dates or by
indexicals like 'now'. Present existence is thus represented by a
two-place predicate: 'Exists (a, now).' Tenses are expressed by
using a later-than predicate, quantifying over times and giving
other predicates an extra argument-place for a time; thus 'a was G'
is analysed as '(3t) (Later (now, t) & G(a,t))'. On Wilson's view,
present-tensed observational sentences get simple analyses: they

temporal operators. Even if we took an interpretation to represent an

instantaneous state of the world, and had temporal operators passing us
between these states as in tense logic, the subject-matter of formulae without
operators would still be spatially local.
Cf. how a Latin letter-writer may use a past tense to describe events
contemporaneous with his act of writing, cf. Cic, Fam., v, 12.2.
170 J. BUTTERFIELD:

have no quantification on the argument-place for times, which is

filled with 'now'. 1
Secondly, there are obvious difficulties in arguing from Sections
2 and 3 for the quantifier analysis of present existence, or for
temporal operators. I will consider here only the case of the
quantifier; my (1984a) considers temporal operators. Even if we
were willing to identify existence (in the sense analysed by the
quantifier) with some kind of observability, Sections 2 and 3 would
not establish that all and only presently existing objects are in some
sense observable. Section 2 shows only that most of the objects
which we each observe presently exist, since most will not have been

Downloaded from http://mind.oxfordjournals.org/ by guest on May 14, 2012

destroyed during the observational time-lag. A few may well have
been destroyed, and even if they have all survived, none of us can
observe all presently-existing objects. Indeed, some presently-
existing objects are observed by nobody. So even if we pool our
observations—i.e. take the quantifier to analyse present observ-
ability by some observer or other—not every presently existing
object will occur in the domain of quantification.

5. Time- Variable and Space- Variable Truth- Value

As Dummett says (1973, p. 390), 'the notion of truth-value as
changing from time to time has a much more powerful appeal to us
than that of truth-value as changing from place to place'. Why?
After all, time-variable and space-variable truth-values are ap-
parently analogous: both can be introduced by an indexical term
('last year's boat race', 'the bridge north of here') or by an indexical
predicate ('was a fiasco', 'is five miles away'). So why does time-
variable truth-value occur more often, or more prominently, than
space-variable truth-value?
I think there are two reasons. The first is language-dependent
and has nothing to do with Sections 2 and 3. It is roughly, that the
grammars of English and many other languages often make time-
variable but not space-variable truth-value compulsory. Thus,
apart from a few cases like 'A stitch in time saves nine' and '2 and 2 is

Quine's better-known (i960, Section 36) analysis is similar. But the

difference—that for Quine there are terms like 'a-at-t' standing for temporal
parts of objects—means that Quine cannot analyse relations between simul-
taneous temporal parts any more simply than those between non-simultaneous
ones. For different terms in a sentence can have different temporal suffixes, e.g.
' fita-at-t, b-at-t')'.
SEEING THE PRESENT 171

4', verbs in English are tensed: they convey information about the
relation of the time spoken about to the time of utterance. (Similarly
for other Indo-European languages.) Since almost all sentences
contain verbs, most of them have a time-variable truth-value. On
the other hand, there is no rule that English verbs, or sentences,
must convey information about the spatial relation of the objects
mentioned to the place of utterance. So unless they happen to
contain a spatially indexical term ('the bridge north of here') or
predicate ('is five miles away'), they will not have a space-variable
truth-value.

Downloaded from http://mind.oxfordjournals.org/ by guest on May 14, 2012

This explanation does not of course depend on verbal tensing—
though some detensers' remarks suggest it does (e.g. Quine 1976, p.
147). It applies to any language in which most sentences relate the
time spoken about to the time of utterance, but not the place spoken
about to the place of utterance. Whether this information is
conveyed as in English, by verbal inflections and auxiliaries, or by
adverbs like 'soon', 'now' and 'recently', or by adverbial phrases
like 'in a while' and 'five days ago', is quite immaterial. 1 But the
explanation will not apply to languages in which spatial indexicals
are at least as frequent as temporal indexicals. The Amerindian
language Kwakiutl is like this. Kwakiutl verbs are indeed tensed;
but its nouns also have compulsory prefixes which convey whether
the object mentioned is near the speaker, or near the audience or
near some other person spoken of, and whether it is visible or
invisible. (These prefixes also serve as pronouns: thus the third
person pronoun 'he' has 6 ( = 3 X 2) basic forms to represent who he
is near and whether he is visible; cf. Boas 1911, pp. 445-446,
527-529). So Kwakiutl speakers might find space-variable truth-
value as 'appealing' as time-variable truth-value.
But they might not. For there is another reason for the greater
appeal of time-variable truth-value, which is quite independent of
language. It relates to observational reports of the kind we can make
before learning the use of watches, maps etc. In making these
reports, we can much more readily specify the objects reported on
without using spatial indexicals then we can specify non-indexically

Chinese and Finnish, which are sometimes cited as tenseless languages, in fact
have some verbal tensing (cf. Chao 1948, pp. 54-55; Olli 1958, pp. 11, 25 ff., 79
ff.). So far as I can judge, most Chinese and Finnish sentences relate the time
spoken about to the time of utterance (by some combination of auxiliaries and
adverbs), but do not relate the objects mentioned to the place of utterance; so
this explanation applies to them.
172 J. BUTTERFIELD:

(e.g. without reference to 'now') when they have the reported

properties and relations. Thus we can often avoid using spatial
indexicals, but we rarely avoid using a temporal indexical like 'now'
or a verbal present tense.
This claim does not of course conflict with the existence of a
language like Kwakiutl. That we can often specify an object without
spatial indexicals does not mean that we have no spatially indexical
information about it, e.g. whether it is near the speaker or the
audience. We do have such information, and can incorporate it into
our observational report if our language, like Kwakiutl, requires us

Downloaded from http://mind.oxfordjournals.org/ by guest on May 14, 2012

to do so. The claim would only conflict with a language which
required observational reports to specify non-indexically the time
for which a predicate is ascribed. For it suggests that unless
speakers of such a language were supplied with watches etc. they
would often be tongue-tied, too ignorant to speak—a hardly
believable state of affairs. At any rate, I know of no such language.
The reason why we can usually avoid spatial indexicals is that
there is usually enough qualitative variation across space. That is,
we can usually distinguish the objects we mention, at least from
other currently observable objects, by their qualities—or at least by
their qualities and their non-indexical spatial relation to other
objects so distinguished.
On the other hand, watches etc. aside, we usually cannot specify
the time non-indexically because the natural 'watches' provided by
daily and seasonal variations are too crude. Using them we can only
describe the present time as, for example, during early morning on a
summer day; and that is imprecise compared with an indexical like
'now'. The latitude allowed by such descriptions is at least an
hour—far greater than the period usually required to make an
observation and communicate it to one's audience, i.e. a few
seconds. So such descriptions cannot generally be substituted in
observational reports for an indexical 'now', or a present tense,
without a crucial loss of information -crucial because the observed
objects will like as not change in the course of an hour.
Of course, the world might have had much less qualitative
variation across space, and might have had natural 'watches' by
which we could describe the present time with a latitude small
enough for the descriptions to be useful in observational reports.
One can imagine a world with objects, changing colour continu-
ously and in step with one another, and common enough for
everyone to be able to see one. But although this is imaginable, there
SEEING THE PRESENT 173

are, as we have already seen in Section 2, physical reasons for not

expecting such frequent changes in objects' observable properties.
So though this reason for the prevalence of temporal indexicals and
thus of time-variable truth-value is contingent, the contingency is
not surprising. 1

6. Sharing a Now but Not a Here

We tend to think that we share a common, albeit ever-changing, now
while we each have a different here. Since we generally want some

Downloaded from http://mind.oxfordjournals.org/ by guest on May 14, 2012

connection between the objectivity of a notion and different
individuals' agreement on it, this intuition apparently supports the
tenser's claim that the present, but not the local, is objective.
However, the detenser can explain this intuition by using Sections 2
and 3 to explicate the metaphorical ideas of sharing a now, and
sharing a here.
The rapidity of oral communication (Section 3) suggests that
people share a now but not a here, in that their interpretations of
tokens of 'now' but not 'here' agree. That is, each person can usually
take a token of 'now' which they hear to refer to the time of
reception, as well as the time of utterance, without misinterpreting
the speaker. But they cannot analogously take a token of 'here' to
refer to where they themselves are.
Explicated like this, a detenser can obviously endorse the
intuition. But is this all there is to it? Suppose that communication
were much slower than it actually is; suppose for example that we
communicated entirely by letter and not at all by speech, gestures
etc. Wouldn't we still have the sense of sharing a now but not a here ?
I think we would; but nevertheless the detenser can still explain the
intuition, though now in terms of observation rather than com-
munication. But this explanation is not an endorsement: if one
explicates sharing a now or here in terms of observation, then
people share a here in strict analogy with their sharing a now.
Rather the detenser explains the intuition by explaining why we

I admit that to get the idea across, I have expressed this reason in a language-
dependent way. I have talked of specifying objects with spatial indexicals and
times with temporal indexicals; and this shows a bias against temporally
indexical specifications of objects and spatially indexical predicates. To put the
point in a language-independent way, think of a report as requiring a property
of some spatial region for some interval of time; the point is then that we can
more often specify the region non-indexically than the interval.
174 J- BUTTERFIELD:

tend to see only one half of this analogy, viz. the sense in which
people share a now.
The obvious way to explicate a common now without appeal to
communication is to say that two people share a now if they agree in
their judgments about what is now the case, i.e. in their present-
tensed judgments. However this suggestion needs clarifying on two
counts; and doing this will lead us to consider observational
judgments. Firstly, if present-tensed judgments are made at
different times they cannot be expected to agree. So we must
apparently fix the reference of the 'now', and explicate sharing a

Downloaded from http://mind.oxfordjournals.org/ by guest on May 14, 2012

now at a time. Secondly, fixing the now should not be sufficient for
agreement; two people can surely share a now while having
disagreements, provided these disagreements can be explained by
their different theoretical beliefs, past experiences, spatial perspec-
tives etc. Presumably these factors are set aside in the case of
ascriptions of observational predicates, other than spatially in-
dexical predicates like 'is far away', to objects that both people can
observe and that are not specified in a spatially indexical way.
Putting these two points together, let us say that two people share
a now at a time if they then make the same ascriptions of such
predicates to such objects. Section 2 shows that people with reliable
perceptions usually do share a now in this sense, for different
people's time-lags in the process of arriving at such ascriptions are
roughly equal. But this is of course a contingent fact; the time-lag
might have been very different for different people, say because
some people were as slow-witted in their observational judgments
as dinosaurs—in which case we would not share a now in this sense.
This explication of sharing a now involves agreement of simul-
taneous observational judgments without spatial indexicals about
what is now the case. The analogous explication of sharing a here
therefore requires agreement of observational judgments made at
the same place without temporal indexicals, about what is here the
case. And on this explication, people do share a here at a given place,
in strict analogy with their sharing a now at a given time: two people
with reliable perceptions will make the same observational judg-
ments of the form 'a is here G at time t' provided they make them at
the same place. (This again is a contingent fact. If people were
slower-witted in their judgments about where they are, they would
sometimes use 'here' to refer to a place they had been rather than to
where they were; and so judgments made in the same place about
what is here the case would sometimes disagree.) So the detenser
SEEING THE PRESENT 175

must explain why we do not see this analogy but instead think of
ourselves as sharing a now but not a here.
Fortunately, the preceding discussion suggests three reasons for
this; the first two come from Section 5, and the third from Section 4.
And the detenser can rely on any or all of them, as he likes; I believe
all three. Firstly, since our language is tensed we take 'sharing a
here' to mean now sharing a here. And people do not usually now
share a here, if we explicate this on analogy with their now sharing a
now, i.e. if we require agreement in their simultaneous judgments
about what is here. That is, let us say that two people share a here at
a time if they then make the same observational judgments about

Downloaded from http://mind.oxfordjournals.org/ by guest on May 14, 2012

what is here. (As before, we consider only observational judgments,
so as to set aside disagreements due to different theoretical beliefs,
past experiences etc.) In this sense, people do not usually share a
here; even if they are close enough to observe the objects located at
each other, they are usually far enough apart for these to be two
different objects, with in general different properties.
Also, the sense in which we share a here (at a given place) is
recherche, for two reasons. First, this sense involves observational
judgments like 'a is here G at t' that have spatial but not temporal
indexicals. And once we set aside devices like watches, we can avoid
temporal indexicals in this way much less often that we can avoid
spatial indexicals, as required in the judgments 'a is now G' etc.
associated with our sharing a now. Second, the sense in which we
share a here is recherche on account of Section 4's point that the
rapidity of oral communication gives us pragmatic reasons for
making present-tensed observational reports. Only if the time-lags
in oral communication were much longer or more variable would
we follow our practice with letters and use a watch etc. to specify
non-indexically the time spoken about; and in that case, judgments
like 'a is here G at t' would now be a recherche as they are. 1

References
Austin, J. 1962 Sense and Sensibilia (Oxford: Clarendon).
Boas, F. 1911 Handbook of American Indian Languages, vol. 1 (Washington:
Smithsonian).
Butterfield, J. 1984 Spatial and Temporal Parts, forthcoming; 1984a Dummett on
Temporal Operators, forthcoming in Philosophical Quarterly.
1 I would like to thank Hugh Mellor for comments on earlier versions; and
Jonathan Grudin, Terry Moore and Stephen Levinson for discussion of the
psychology and linguistics literature. Thanks also to audiences at talks in
Chicago, Cambridge and London.
176 J. BUTTERFIELD: SEEING THE PRESENT

Chao, Y. 1948 A Mandarin Primer (Cambridge Mass.: Harvard).

Dummett, M. i960 A Defence of McTaggart's Proof of the Unreality of Time.
Philosophical Review 69, pp. 497-504; 1968 The Reality of the Past,
Proceedings of the Aristotelian Society 69, pp. 239-258. 1973 Frege: Philosophy
of Language (London: Duckworth).
Flowers, J. 1975 Sensory Interference in a Word-color Watching Task, Perception
and Psychophysics 18, pp. 37-43.
Gale, R. 1964 Is it now now?, Mind 73, pp. 97-105; 1968 The Language of Time
(London: Routledge and Kegan Paul).
Garson, J. 1971 Here and Now, in Basic Issues in the Philosophy of Time, E.
Freeman and W. Sellars eds. (La Salle: Open Court), pp. 145-154. Also in
Monist S3 (1969), pp. 469-477.
Lloyd, G. 1978 Time and Existence, Philosophy 53, pp. 215-228.
Mellor, D. 1981 Real Time (Cambridge: Cambridge University Press).

Downloaded from http://mind.oxfordjournals.org/ by guest on May 14, 2012

Olli, J. 1958 Fundamentals of Finnish Grammar (New York).
Posner, M. and Mitchell, R. 1967 Chronometric Analysis of Classification,
Psychological Review 74, pp. 392—409.
Potter, M. and Faulconer, B. 1975 Time to Understand Pictures and Words,
Nature 253, pp. 437-438.
Prior, A. 1959 Thank Goodness That's Over, Philosophy 34, pp. 12-17. 1970 The
Notion of the Present, in The Study of Time]. Fraser ed. (Berlin: Springer).
Putnam, H. 1967 Time and Physical Geometry, Journal of Philosophy 64, pp.
240-247.
Quine, W. i960 Word and Object (Cambridge: MIT). 1976 Ways of Paradox
(Cambridge Mass.: Harvard).
Shaffer, W. and LaBerge, D. 1979 Automatic Semantic Processing of Unattended
Words, Journal of Verbal Learning and Verbal Behaviour 18, pp. 413-426.
Taylor, R. 1955 Spatial and Temporal Analogies and the Concept of Identity,
Journal of Philosophy 52, pp. 599-612.
Williams, B. 1981 Another time, another place, another person, in Moral Luck
(Cambridge: University Press), pp. 164—173.
Wilson, N. 1955 Space, Time and Individuals, Journal of Philosophy 52, pp.
589-598.
Zemach, E. 1972 'Here' and 'Now', Mind 81, pp. 251-255.

JESUS COLLEGE,
C A M B R I D G E , CB5 8BL